Research Areas

Epidemiology

These models simulate how an outbreak moves through a real population, either region by region or person by person. As surveillance data comes in, the simulations are tuned to match it. They then answer two different questions: what is likely to happen over the next few weeks, and what would happen under a given policy. Signals beyond case counts help sharpen the picture, including virus levels measured in wastewater. Much of this work ran live, feeding CDC FluSight and the national COVID-19 forecasting and scenario hubs.

Publications in this area →

Evolution & ecology

Whatever is spreading is usually changing as it spreads. For pathogens, that means models in which new variants compete and one can sweep through a population. The practical question is how much genetic sequencing is enough to catch a rising variant early. The same tools carry over to invasive pests and weeds, which travel along trade routes for food and other goods rather than through people. Satellite imagery and habitat models then map where a species like the tomato leafminer is likely to take hold next.

Publications in this area →

Migration & mobility

Where people go determines what spreads and who needs help. This work reconstructs movement at every scale, from daily commutes to international travel, drawing on anonymized location data, travel surveys, and census records. Those patterns then feed models of how a virus travels between places. A second strand studies displacement itself. Simulations developed around the war in Ukraine follow how people fleeing conflict choose routes, pick destinations, and eventually return. A running question is how much geographic detail a forecast actually needs.

Publications in this area →

Social & information networks

Ideas and behaviors spread along social ties in ways that look a lot like infection. The shape of the network decides what catches on. This thread follows how influence travels, how posts and stories compete for a limited pool of attention, and how research collaborations form and shift over time. The link back to public health is direct. Knowing who is connected to whom tells you where a limited supply of vaccine does the most good, rather than where age or region alone would send it.

Publications in this area →

Methods & machine learning

This is the engine room, where simulations are detailed enough to give every person in a country their own daily schedule. Running thousands of them at once takes serious computing. On top of that sits a statistical layer. It blends several imperfect models into one forecast that beats any of them alone, then checks whether the stated confidence holds up in practice. Being straight about what is still uncertain is part of the job. Recent work adds shared benchmarks for comparing methods, and AI assistants that help set up and interpret the models.

Publications in this area →

Areas are assigned from the language of the abstracts themselves, using the same themes as the word cloud on the home page, so a paper spanning several areas appears under each of them. A handful of older papers do not match any area and are visible only in the full publications list.